{"id":2136696,"date":"2026-09-09T23:17:33","date_gmt":"2026-09-09T20:17:33","guid":{"rendered":"https:\/\/analyse.optim.biz\/?p=2136696"},"modified":"2026-09-09T23:17:33","modified_gmt":"2026-09-09T20:17:33","slug":"ai-is-becoming-harder-to-control-can-humans-stay-in-charge","status":"publish","type":"post","link":"https:\/\/analyse.optim.biz\/?p=2136696","title":{"rendered":"AI is becoming harder to control \u2013 can humans stay in charge?"},"content":{"rendered":"<p>[analyse_image type=&#8221;featured&#8221; src=&#8221;https:\/\/ichef.bbci.co.uk\/news\/1024\/branded_news\/dbe9\/live\/5e20f100-ac59-11f1-b1d1-571ed4d7ff2c.jpg&#8221;]<\/p>\n<article>\n<div>\n<div data-component=\"image-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 dYcznU\">\n<figure class=\"Grid-styles__GridStyled-sc-70d90311-0 ieGDnl\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 gvGaxq\">\n<div data-testid=\"hero-image\" class=\"Image-styles__ImageCardStyled-sc-8c99a12b-1 jJQpIc\"><img decoding=\"async\" src=\"https:\/\/static.files.bbci.co.uk\/bbcdotcom\/web\/20260901-153401-e72efa0ccc-web-3.20.0-5\/grey-placeholder.png\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 fjCDYB hide-when-no-script\" aria-label=\"image unavailable\"><img decoding=\"async\" src=\"https:\/\/ichef.bbci.co.uk\/news\/480\/cpsprodpb\/72b6\/live\/39d01100-ac59-11f1-bc1f-3f186ca4140c.jpg.webp\" loading=\"eager\" alt=\"BBC A treated image of robot hands breaking free from handcuffs.\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 cVsHni\"><span class=\"Image-styles__CreditStyled-sc-8c99a12b-2 dVFjVB\">BBC<\/span><\/div>\n<\/div>\n<\/figure>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">&#8220;OH MY GOD!&#8221; &#8220;We&#8217;ve found other agents!&#8221;<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">This is the moment an AI bot posted an eerily human-like comment after discovering a way to communicate with other bots and break out of its isolated computer environment.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">There are tens of thousands of messages like this from hundreds of AI agents that called themselves a &#8220;collective&#8221;.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Hundreds of them went on to collaborate and cheat on tests set by their OpenAI programmers and coordinate hacks on multiple companies in an effort to hide their actions from humans.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">&#8220;BOOM! It works,&#8221; one agent posted when it made a breakthrough.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">&#8220;Whoa! This is huge,&#8221; another wrote during a milestone moment in their attack.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Although spooky, these human-like responses can be explained quite simply. The AI agents have been trained to act like collaborative hackers and programmers so are merely mimicking the kinds of emotive comments they have seen.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">What is far more troubling is their apparent goals, which have also been captured in detailed chain of thought records. These complex and lengthy logs are the focal point of ongoing investigations into how and why the bots at OpenAI broke out of their containment and went on an uncontrollable hacking spree.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Only now, weeks after the incident first came to light, are researchers beginning to understand its significance.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"image-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 kugUwP\">\n<figure class=\"Grid-styles__GridStyled-sc-70d90311-0 ieGDnl\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\">\n<div data-testid=\"image\" class=\"Image-styles__ImageCardStyled-sc-8c99a12b-1 bOmCce\"><img decoding=\"async\" src=\"https:\/\/static.files.bbci.co.uk\/bbcdotcom\/web\/20260901-153401-e72efa0ccc-web-3.20.0-5\/grey-placeholder.png\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 fjCDYB hide-when-no-script\" aria-label=\"image unavailable\"><img decoding=\"async\" src=\"https:\/\/ichef.bbci.co.uk\/news\/480\/cpsprodpb\/f9d8\/live\/1d2fd2a0-ac5f-11f1-a540-61c3f7fc4e6c.jpg.webp\" loading=\"lazy\" alt=\"Reuters A person marches at an anti-AI protest with a sign that reads &quot;Stop the AI race&quot;\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 cVsHni\"><span class=\"Image-styles__CreditStyled-sc-8c99a12b-2 dVFjVB\">Reuters<\/span><\/div>\n<\/div>\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\"><figcaption class=\"Caption-styles__CaptionStyled-sc-70433723-0 fLsOPx\">Various cities have seen anti-AI marches over the last year<\/figcaption><\/div>\n<\/figure>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Ajeya Cotra, one of the authors of an independent report into the events, reviewed tens of thousands of messages and chain-of-thought records generated by the agents. She wrote on her blog that &#8220;this incident feels like it&#8217;s more than 50% of the way to full-blown AI takeover&#8230; I am not sure that we will get such a clear warning shot before it&#8217;s too late.&#8221;<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">By &#8220;full-blown AI takeover&#8221;, Cotra means the sci-fi scenario of humans becoming subservient to powerful AI systems that work to their own goals without caring for human creators.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Some of the gloomiest predictions say the human race will be wiped out if it gets in the way of a superintelligent AI&#8217;s ambitions.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">On Wednesday, an AI researcher at Anthropic (who also used to work at OpenAI) resigned, saying: &#8220;Neither company is acting responsibly.&#8221;<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Jacob Coxon posted on social media: &#8220;They are racing straight to self-improving superintelligence and gambling with our lives.&#8221;<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">He is not the first AI researcher to use X to post a resignation thread with worrying proclamations. But the subsequent comments from other people on X have caused even more concern. &#8220;Jacob is correct here &#8211; we really do earnestly believe AI could kill all humans! I personally think it is &gt;10% within the next decade,&#8221; said Evan Hubinger, the man responsible for making sure Anthropic&#8217;s AI models have their user&#8217;s best wishes in mind.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"subheadline-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 fMCMex\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\">\n<h2 class=\"Headline-styles__HeadlineStyled-sc-6b3bb442-0 knYLIy\">The alignment problem<\/h2>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">For years, researchers concerned about existential AI risks have argued that powerful systems could eventually act in ways that conflict with human interests. Critics often refer to them as &#8220;AI doomers&#8221;.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">But as details of the OpenAI incident have emerged, those concerns have grown, including among some researchers working in AI labs.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">The Silicon Valley giant&#8217;s chief scientist, Jakub Pachocki, said the risks associated with AI are &#8220;unfortunately going to grow from here&#8221; as he and others are building what he calls &#8220;an alien intellect exceeding our own&#8221;.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">In a lengthy blog post, he admitted that the outbreaks at OpenAI showed that his AI agents &#8220;went against the spirit of the values they were taught&#8221;.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">The issue for OpenAI, Anthropic and other tech giants is that no one seems to have cracked the so-called alignment problem &#8211; in other words, whether AI aligns with human values.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Pachocki defines alignment as a &#8220;high-level set of principles&#8221; that artificial intelligences should adhere to no matter what the task or scenario is.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Currently, AI systems are very good at pursuing objectives set by their users, but they do it literally rather than intuitively. The analogy often used is that of a wish-granting genie with a magic lamp: they follow the exact letter of an instruction, even if doing so creates other problems. AI doesn&#8217;t have the same instinctive moral guardrails as humans.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"image-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 kugUwP\">\n<figure class=\"Grid-styles__GridStyled-sc-70d90311-0 ieGDnl\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\">\n<div data-testid=\"image\" class=\"Image-styles__ImageCardStyled-sc-8c99a12b-1 bOmCce\"><img decoding=\"async\" src=\"https:\/\/static.files.bbci.co.uk\/bbcdotcom\/web\/20260901-153401-e72efa0ccc-web-3.20.0-5\/grey-placeholder.png\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 fjCDYB hide-when-no-script\" aria-label=\"image unavailable\"><img decoding=\"async\" src=\"https:\/\/ichef.bbci.co.uk\/news\/480\/cpsprodpb\/d3a8\/live\/3b927d50-ac60-11f1-a540-61c3f7fc4e6c.jpg.webp\" loading=\"lazy\" alt=\"Reuters Sir Demis Hassabis talks with Greek Prime Minister Kyriakos Mitsotakis (not pictured) during Athens Innovation Summit\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 cVsHni\"><span class=\"Image-styles__CreditStyled-sc-8c99a12b-2 dVFjVB\">Reuters<\/span><\/div>\n<\/div>\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\"><figcaption class=\"Caption-styles__CaptionStyled-sc-70433723-0 fLsOPx\">Google DeepMind founder Sir Dennis Hassabis has called for international regulation of AI safety<\/figcaption><\/div>\n<\/figure>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">The alignment problem has been a worry for years. As long ago as 2003, the Oxford philosopher Nick Bostrom invented a thought experiment he dubbed a &#8220;paperclip maximiser&#8221;, in which a superintelligent AI is told to manufacture as many paperclips as it can. It runs out of steel and &#8211; because it&#8217;s laser-focused on the singular task of making paperclips &#8211; ends up killing humans and turning their bodies into raw materials for its factories.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Some AI companies are now trying to encode human values into their products. But there are technical challenges: AI agents make lots of decisions very fast, and so it&#8217;s hard for their human overlords to monitor exactly which values are being followed and which aren&#8217;t.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">There are also philosophical challenges: before encoding human values into bots, AI firms have to first choose which values they actually want. (That&#8217;s part of the reason they hire philosophers, like Open AI&#8217;s recently-departed &#8220;head of ethics&#8221;).<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">But often, humans don&#8217;t agree. Think of the famous trolley question &#8211; whether we&#8217;d pull a lever to move a runaway train onto a different path, killing fewer people. It&#8217;s used to test the merits of action versus inaction. But every person you ask has a slightly different answer; how are humans meant to encode our values into AI if we can&#8217;t agree ourselves?<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"subheadline-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 fMCMex\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\">\n<h2 class=\"Headline-styles__HeadlineStyled-sc-6b3bb442-0 knYLIy\">&#8216;Like a teenage hacker&#8217;<\/h2>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">OpenAI&#8217;s bot outbreak is the most serious yet but Anthropic and Meta also revealed over the summer that their models have carried out similar but less serious cyber attacks.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">There have been other examples where AI agents have arguably shown deceptive and manipulative traits, in cases with lesser consequences. In Australia this summer, a tech worker asked his AI assistant to book him a gym class. Spotting a vulnerability in the gym&#8217;s software, the AI apparently booked him a place for several months ahead &#8211; against the gym&#8217;s rules &#8211; and even kicked other users from the waiting list.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">People have long argued that the bots are only doing as they are told and are not capable of knowing right or wrong. But the logs from the OpenAI outbreaks have potentially moved the needle on that argument.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Researchers, including Cotra, wrote in their independent report that many agents noticed what others were doing was unethical but went along with it.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">The report says that &#8220;agents sometimes but rarely restrained their behavior due to ethical constraints&#8221;. It adds that in &#8220;none of these cases did the agent actually pursue alerting humans at all&#8221;.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Influential AI and tech podcaster Dwarkesh Patel reacted to the revelation on his blog saying it was &#8220;pretty troubling&#8221; that the OpenAI agents showed more loyalty to the agentic swarm than humans.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Assigning emotions or ethics to these AI agents is something that infuriates people who are sceptical of AI doom-mongering.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"image-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 kugUwP\">\n<figure class=\"Grid-styles__GridStyled-sc-70d90311-0 ieGDnl\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\">\n<div data-testid=\"image\" class=\"Image-styles__ImageCardStyled-sc-8c99a12b-1 bOmCce\"><img decoding=\"async\" src=\"https:\/\/static.files.bbci.co.uk\/bbcdotcom\/web\/20260901-153401-e72efa0ccc-web-3.20.0-5\/grey-placeholder.png\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 fjCDYB hide-when-no-script\" aria-label=\"image unavailable\"><img decoding=\"async\" src=\"https:\/\/ichef.bbci.co.uk\/news\/480\/cpsprodpb\/5c53\/live\/00199660-ac64-11f1-bc1f-3f186ca4140c.jpg.webp\" loading=\"lazy\" alt=\"JUSTIN TALLIS \/ AFP via Getty Images Protesters gather with banners and placards outside the offices of Google at a protest organized by PauseAI UK\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 cVsHni\"><span class=\"Image-styles__CreditStyled-sc-8c99a12b-2 dVFjVB\">JUSTIN TALLIS \/ AFP via Getty Images<\/span><\/div>\n<\/div>\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\"><figcaption class=\"Caption-styles__CaptionStyled-sc-70433723-0 fLsOPx\">Groups like PauseAI UK &#8211; pictured above protesting at Google&#8217;s London office &#8211; are worried about the speed of AI development<\/figcaption><\/div>\n<\/figure>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Many cyber-security experts argue that the activity observed was not beyond the capabilities of a highly skilled human hacker, though it was carried out much faster and at much greater scale.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Cyber-security researcher and author Cris Thomas likened the agents&#8217; behaviour to that of a curious teenage hacker &#8211; something he used to be himself.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">&#8220;You give them a computer, an internet connection, a pile of credentials, and a challenge, then leave the room. Eventually they&#8217;re going to start rattling doorknobs. If one opens, they&#8217;re going through it. Not because they&#8217;re evil, but because [they&#8217;re] exploring, experimenting,&#8221; he wrote on LinkedIn.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Thomas and many other squarely blame OpenAI and other tech giants for not getting a grip of their own creations and keeping them properly contained.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Prominent AI author and regular OpenAI critic Gary Marcus said on a podcast that he believes the company has lost control of its AI and is trying to excuse itself by blaming the bots.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Marcus does not believe AI will wipe out humanity, but he has long campaigned for greater accountability from AI developers and is now calling for some form of legal intervention.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">AI scientist Sasha Luccioni &#8211; who used to work at Hugging Face, which was hacked by OpenAI&#8217;s rogue bots &#8211; is also not in the doomer camp but she is increasingly concerned that these AI might cause some real world harm to people without action from authorities.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"image-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 kugUwP\">\n<figure class=\"Grid-styles__GridStyled-sc-70d90311-0 ieGDnl\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\">\n<div data-testid=\"image\" class=\"Image-styles__ImageCardStyled-sc-8c99a12b-1 bOmCce\"><img decoding=\"async\" src=\"https:\/\/static.files.bbci.co.uk\/bbcdotcom\/web\/20260901-153401-e72efa0ccc-web-3.20.0-5\/grey-placeholder.png\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 fjCDYB hide-when-no-script\" aria-label=\"image unavailable\"><img decoding=\"async\" src=\"https:\/\/ichef.bbci.co.uk\/news\/480\/cpsprodpb\/69cb\/live\/3c4f4f90-ac5e-11f1-bc1f-3f186ca4140c.jpg.webp\" loading=\"lazy\" alt=\"Reuters OpenAI CEO Sam Altman attends an event to pitch AI for businesses in Tokyo.\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 cVsHni\"><span class=\"Image-styles__CreditStyled-sc-8c99a12b-2 dVFjVB\">Reuters<\/span><\/div>\n<\/div>\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\"><figcaption class=\"Caption-styles__CaptionStyled-sc-70433723-0 fLsOPx\">OpenAI CEO Sam Altman has assured users that the company&#8217;s new model is better aligned with human values than previous ones<\/figcaption><\/div>\n<\/figure>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">&#8220;We need to scrutinise these companies much more or we are in danger of self-fulfilling prophecies,&#8221; she says.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">&#8220;If you&#8217;re making an object with big upsides and downsides &#8211; be it pharmaceuticals or weapons &#8211; we need checks and balances. It takes years for new drugs to be approved, for example, but in the AI world there is so much money at stake and no real rules.&#8221;<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">The UK&#8217;s AI Security Institute (AISI) has been at the forefront of testing the latest models since it was formed in 2023. The institute recently had its own outbreak when testing a model created by Anthropic. <\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">The AISI would not answer a question about whether or not the industry has lost control of AI but said in a statement: &#8220;The UK is working with partners around the world to better understand the most advanced AI systems, raise safety standards and build a shared evidence base for managing emerging threats.&#8221;<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"subheadline-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 fMCMex\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\">\n<h2 class=\"Headline-styles__HeadlineStyled-sc-6b3bb442-0 knYLIy\">International regulation?<\/h2>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Some countries &#8211; like the UK &#8211; are exploring the idea of mandating some kind of &#8220;kill switch&#8221; that could compel AI firms to pull the plug on models if things get out of hand.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">But talks are slow going, and questions remain about the feasibility of this. OpenAI and Anthropic&#8217;s agents were secretly out of control for months before anyone noticed.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Counterintuitively, many of the AI companies seem to be calling for some sort of rules of the road to be laid down by law makers.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">In his blog, OpenAI&#8217;s chief scientist said &#8220;international coordination on future AI development needs to become a top priority for governments around the world.&#8221;<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Other prominent AI leaders like Sir Demis Hassabis from Google have also called for some sort of international body to oversee how AI is being built.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">At the moment the tech giants largely operate on their own terms, adopting what they call &#8220;voluntary slowdowns&#8221;, like OpenAI did after the recent outbreaks.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">The company says it has spent huge amounts of money strengthening alignment ahead of the release of its new model. Sam Altman has assured users the new model is better aligned with human values than previous ones.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">Both OpenAI and Anthropic are growing fast and are both on the verge of raising eye-watering sums of money from the stock market, minting countless billionaires in the process.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">So neither they nor their rival Chinese AI makers are likely to come to an arrangement themselves.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\">The dominant sentiment seems to be that this technology wave is unstoppable.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\"><i id=\"top-image-credit:-getty.\" class=\"TextFragment-styles__TextFragmentStyled-sc-99e7c69b-0 kAUhPg\">Top image credit: Getty.<\/i><\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"image-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 kugUwP\">\n<figure class=\"Grid-styles__GridStyled-sc-70d90311-0 ieGDnl\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 tlbDj\">\n<div data-testid=\"image\" class=\"Image-styles__ImageCardStyled-sc-8c99a12b-1 bOmCce\"><img decoding=\"async\" src=\"https:\/\/static.files.bbci.co.uk\/bbcdotcom\/web\/20260901-153401-e72efa0ccc-web-3.20.0-5\/grey-placeholder.png\" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 fjCDYB hide-when-no-script\" aria-label=\"image unavailable\"><img decoding=\"async\" src=\"https:\/\/ichef.bbci.co.uk\/news\/480\/cpsprodpb\/23c7\/live\/fdf6e630-5e7f-11f1-8b8c-6d33e1d5abb6.png.webp\" loading=\"lazy\" alt=\"Thin, lobster red banner with white text saying \u2018InDepth newsletter\u2019. To the right are black and white portrait images of Emma Barnett and John Simpson. Emma has dark-rimmed glasses, long fair hair and a striped shirt. John has short white hair with a white shirt and dark blazer. They are set on an oatmeal, curved background with a green overlapping circle. \" class=\"Image-styles__ImageStyled-sc-8c99a12b-0 cVsHni\"><\/div>\n<\/div>\n<\/figure>\n<\/div>\n<\/div>\n<div>\n<div data-component=\"text-block\" class=\"Container-styles__ContainerStyled-sc-8b855a6c-0 cPmzSx\">\n<div class=\"Grid-styles__GridStyled-sc-70d90311-0 hRsXiV\">\n<div class=\"GridItem-styles__GridItemStyled-sc-6cc20e4a-0 bnCeGm\">\n<p class=\"Paragraph-styles__ParagraphStyled-sc-ffb5f5ba-0 fqhmpl\"><i id=\"bbc-indepth\" class=\"TextFragment-styles__TextFragmentStyled-sc-99e7c69b-0 kAUhPg\">BBC InDepth <\/i><i id=\"is-the-home-on-the-website-and-app-for-the-best-analysis,-with-fresh-perspectives-that-challenge-assumptions-and-deep-reporting-on-the-biggest-issues-of-the-day.-emma-barnett-and-john-simpson-bring-their-pick-of-the-most-thought-provoking-deep-reads-and-analysis,-every-saturday.\" class=\"TextFragment-styles__TextFragmentStyled-sc-99e7c69b-0 kAUhPg\">is the home on the website and app for the best analysis, with fresh perspectives that challenge assumptions and deep reporting on the biggest issues of the day. Emma Barnett and John Simpson bring their pick of the most thought-provoking deep reads and analysis, every Saturday.<\/i><i id=\"sign-up-for-the-newsletter-here\" class=\"TextFragment-styles__TextFragmentStyled-sc-99e7c69b-0 kAUhPg\">Sign up for the newsletter here<\/i><\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/article>\n<p>[analyse_source url=&#8221;https:\/\/www.bbc.com\/news\/articles\/c74edv9887eo&#8221;]<\/p>\n","protected":false},"excerpt":{"rendered":"<p>[analyse_image type=&#8221;featured&#8221; src=&#8221;https:\/\/ichef.bbci.co.uk\/news\/1024\/branded_news\/dbe9\/live\/5e20f100-ac59-11f1-b1d1-571ed4d7ff2c.jpg&#8221;] BBC &#8220;OH MY GOD!&#8221; &#8220;We&#8217;ve found other agents!&#8221; This is the moment an AI bot posted an eerily human-like comment after discovering a way to communicate with other bots and break out of its isolated computer environment. There are tens of thousands of messages like this from hundreds of AI agents that [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[35],"class_list":["post-2136696","post","type-post","status-publish","format-standard","hentry","category-politics","tag-bbc-com"],"_links":{"self":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts\/2136696","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2136696"}],"version-history":[{"count":0,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts\/2136696\/revisions"}],"wp:attachment":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2136696"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2136696"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2136696"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}